(12) unitedstatespatent (10) patentno.: us7,585,356b2 · us007585356b2 (12) unitedstatespatent (10)...
TRANSCRIPT
US007585356B2
(12) United States Patent (10) Patent No.: US 7,585,356 B2
Oyama et al. (45) Date of Patent: Sep. 8, 2009
(54) HYDROTHERMALLY—STABLE OTHER PUBLICATIONS
SILICA-BASED COMPOSITE MEMBRANES
FOR HYDROGEN SEPARATION
(75) Inventors: S. Ted Oyama, Blacksburg, VA (US);
Yungeng Gu, Painted Post, NY (US)
(73) Assignee: Virginia Tech Intellectual Properties,
Inc., Blacksburg, VA (US)
( * ) Notice: Subject to any disclaimer, the term of this
patent is extended or adjusted under 35
U.S.C. 154(b) by 603 days.
(21) Appl.No.: 11/381,088
(22) Filed: May 1, 2006
Prior Publication Data
US 2007/0251388 A1 NOV. 1, 2007
(65)
Int. Cl.
B01D 53/22 (2006.01)
US. Cl. ........................... 96/11; 96/4; 96/7; 96/10;
95/55; 95/56
Field of Classification Search ....................... 96/4,
96/7, 10711; 95/55, 56
See application file for complete search history.
(51)
(52)
(58)
References Cited
U.S. PATENT DOCUMENTS
(56)
8/ 1994 Anderson et a1.
2/1999 Lai et a1. .................... 210/653
12/2000 Kondo et a1.
1/2003 Kulkarni et a1.
3/2003 Oyama et a1.
6,730,364 B2 5/2004 Hong et a1.
6,854,602 B2 2/2005 Oyama et a1.
2003/0183080 A1* 10/2003 Mundschau .................... 95/55
2004/0241520 A1* 12/2004 Ha et a1. ....................... 429/33
5,342,431 A
5,871,650 A *
6,159,542 A
6,508,860 B1
6,527,833 B1
Vent
Oxygen
Trap
Furnace
Balance Gas
H20
Oxygen
& Water
Trap
DiEute Gas
Eversteijn, F. C.; “Low-Temperature Deposition of Alumina-Silica
Films”; Philips Res. Repts 21, 379-386; 1966.
de Vos, Renate M.; Maier, Wilheim F. ; Verweij, Henk; “Hydrophobic
Silica Membranes for Gas Separation”; Journal of Membrance Sci-
ence 158; 1999; 277-288.
Yoshida, Kazuhiro; Hirano, Yoshio; Fujii, Hironori; Tsuru,
Toshinori; Asaeda, Masashi; “Hydrothermal Stability and Perfor-
mance of Silica-Zirconia membranes for Hydrogen Separation in
Hydrothermal Conditions”; Journal of Chemical Engineering of
Japan; 2001; vol. 34, 523-530.
Nam, SukWoo; Ha, HeungYong;Y00n, Sung Pil; Han, Jonghee; Lim,
Tae Hoon; Oh, In-Hwan; Hong, Seong-Ahn; “Hydrogen-Permselec-
tive TiOZ/SiOZ Membranes Formedby Chemical Vapor Deposition”;
Korean Membrane Journal; Dec. 2001; vol. 3 N0. 1; 69-74.
(Continued)
Primary ExamineriRobert J Hill, Jr.
Assistant ExamineriDung Bui
(74) Attorney, Agent, or FirmiBaker & McKenzie LLP
(57) ABSTRACT
Thin layers ofa mixed composition are deposited on a porous
substrate by chemical vapor deposition in an inert atmosphere
at high temperature. The resulting membrane has excellent
stability to water vapor at high temperatures. An exemplary
membrane comprises an amorphous mixed-element surface
layer comprising silica and at least one oxide of additional
element, an optional porous substrate on which said surface
layer is deposited, and a porous support on which said sub-
strate or mixed-element surface layer is deposited, wherein
the permeance of the membrane is higher than 1><10'7 mol
m"2 5'1 Pa"1 and the selectivity ofH2 over C0, C02, and CH4
is larger than 100, and wherein the H2 permeance of the
membrane after exposure to a stream containing 60 mol %
water vapor at 673 K for 120 h is at least 50% of its initial H2
permeance.
13 Claims, 25 Drawing Sheets
Temperature
Conlroller
Membrane
Mass Flow
Controller2—Way
Valve
Oxygen 3
8: Water
Trap
Carrier Gas
Waler Bath
US 7,585,356 B2
Page 2
OTHER PUBLICATIONS
Wu, J.C.S.; Sabol, H.; Smith, G.W.; Flowers, D.L.; Liu, P.K.T.;
“Characterization ofHydrogen-Permselective Microporous Ceramic
Membranes”; Journal of membrane Science 96; 1994; 275-287.
Lin, Y.S.; Kumakiri, 1.; Nair, B.N.; Alsyouri, H.; “Microporous Inor-
ganic Membranes”; Separation and Purification Methods; vol. 31,
No. 2, 2002; 229-379; Marcel Dekker, Inc.; USA.
Sea, Bong-Kuk; Soewito, Eddy; Watanabe, Midori; Kusakabe,
Katsuki; Morooka, Shigeharu; Kim, Sung Soo; “Hydrogen Recovery
from a H2-H20-HBr Mixture Utilizing Silica-Based Membranes at
Elevated Temperatures. 1. Preparation of H20- and H2-Selective
Membranes”; Ind. Eng. chem. Res.; May 1998; 37, 2502-2508.
Fotou, G. P.; Lin, Y.S.; Pratsinis, S.E.; “Hydrothermal Stability of
Pure and Modified Microporous Silica Membranes”; Journal of
Materials Science; 1995; 39, 2803-2808; Chapman & Hall; USA.
Yan, Shenghun; Maeda, Hideaki; Kusakabe, Katsuki; Morroka,
Shigeharu; Akiyama, Yasunobu; “Hydrogen-Permselective SiO2
Membrane Formed in Pores of Alumina Support tube by chemical
Vapor Deposition With Tetraethyl Ortho silicate”; Materials and Inter-
faces; Ind. Eng. Chem. Res.; 1994; 33, 2096-2101.
Hekkink, H.A.; De Lange, R.S.A.; Hoeve, A.A. Ten; Blankenvoorde,
P.J.A.M.; Keizer, K.; Burggraaf, A.J.; “Characterization and Perme-
ation Properties of Binary SiO2-TiO2 and SiO2-Al203 Modified
Gamma-Alumina Membranes”; Key Engineering Materials; 1991;
vols. 61 & 62, 375-378; Copyright Trans Tech Publications, Switzer-
land.
Okubo, Tatsuya; 1noue, Hakuai; “Introduction of Specific Gas Selec-
tivity to Porous Glass Membranes By Treatment With
Tetraethoxysilane”; Journal of Membrane Science; 1989; 109-117;
Elsevier Science Publishers B. V., Amsterdam, The Netherlands.
Gavalas, G.R; Megiris, C.E.; Nam, S.W.; “Deposition of
H2-Permselective SiO2 Films”; Chemical Engineering and Science;
1989; vol. 44, No. 9, 1829-1835; Great Britain.
Masaryk, Joseph S. and Fulrath, Ricth M.; “Diffusivity of Helium
in Fused Silica”; The Journal of chemical Physics; Aug. 1973; vol.
59, No. 3; 1198-1202; USA.
* cited by examiner
U.S. Patent
Sep. 8, 2009 Sheet 1 of 25 US 7,585,356 B2
25 a -:
E5840
20 — D\ —
/ D 88200
o\"
“a. C] C}
«5 15 - \ O/ 0 138630 -
a: AE '3 \ A \
a / 0 ma} To " O \ A A m
g '3 1k: / o \A
“5 / \ \ \> \ / \ A
5 _ / u o 0 \AA _.
\ / 0\ AAA
{:1 Db A CC) 2)
0 _ 0/ BBDGGUBEQQ 000000 00 AAA _
10 100 1000 6000
Particle diameter / nm
Fig. 1
US. Patent Sep. 8, 2009 Sheet 2 of 25 US 7,585,356 B2
mSpeed
"controller
Membrane
tUbB\
Stepping
motor
Dipping
solution
US. Patent Sep. 8, 2009 Sheet 3 of 25 US 7,585,356 B2
[38630
dV/dlog(D)
'1 ' .HIH10 100
Pore diameter/ nm
Fig. 3
US. Patent Sep. 8, 2009 Sheet 4 of 25 US 7,585,356 B2
US. Patent Sep. 8, 2009 Sheet 5 of 25 US 7,585,356 B2
Vent I
Vent 4——
5:: <39
m ml m E! C!
Furnace '1 : Temperature
\ Controller
MFC Membrane
7; .
U . 2 Heating Mass Flow
"'0” 5/ Tape ControllerCross ;
at
7 Balance Gas
MFC —$-———
333551;: 2~way Oil Bath ”W"
.‘ Oxygen......
1:33:35 :3} & Water
::::‘. 2": if Trap
Dilute Gas
Water Bath
Fig. 5
US. Patent Sep. 8, 2009 Sheet 6 of 25 US 7,585,356 B2
1E-4g , . . . E _500
: O
: D wa—HEICHd
film 1E~5§~ Z:
:1 3 —: 100
“:5” 1E-6 -
E i 2::
’5 ‘ IE
3 : ~10 fl
G} :
E 3
Si 159;—
1E-‘10‘ r - - . . 1
Deposition Time/ h
Fig. 6
US. Patent Sep. 8, 2009 Sheet 7 of 25 US 7,585,356 B2
1 Ew4
E I I i a | a I
E O we— HEICHH
iE-S _— WE.” Hzico 3 1000
Tm § “A“ HZICOZ —D- .
*z“ 1E~6 g-
m 5
E 1E5? l— 1 ‘100 r?o s ' ._>_
EE
" g
E '1 E43 : ”(:3o 5 (D
{3 ~
g ' - 10E 1 E-g $— .WQW 3—12 :
115 I g ~m— CH4
iE—iO ~ 3. —~A—— co
I mvm (302
1E—i '1 . . ' - - . - ' - 1
0 1 2 3 4 5
Deposition Time / h
Fig. 7
US. Patent Sep. 8, 2009 Sheet 8 of 25 US 7,585,356 B2
i F I . E ' é ' I ' 2 1 E-5
$090 —: o q
- “in;
a» ‘—\ ' 3.E 0 mm
8 100 1 \ —————~——- E
as s \ — 1E-6 “503¢ E
I ‘N.
ii. 8I 10 /<C:><o E
5/“ E_ o o.
—_ 1E~7 3:“
‘ I n J c l , f r I
0.00 002 0,04 OHDB 0.08 010 0.12
ATSB / TEOS (moiar)
Fig. 8
US. Patent Sep. 8, 2009 Sheet 9 of 25 US 7,585,356 B2
1543
155;. 31000‘Tm g _
CL.
‘7 1E"6§'
“3 5
E 15:7;- “- 100 ’5‘0 3 - .2
E 3 ' g
g :
{D .. E“ ':E 1E9: :10
ES
“- 13-10?
1E—11' . - . - . - - - . - 1
O 1 2 3 4 5
Deposition Time / h
Fig. 9
US. Patent Sep. 8, 2009 Sheet 10 of 25 US 7,585,356 B2
1E“6 _ I ' I . i . I I i
‘T "' D
53I: ATSBITEOS=00065
UJ
WE MVRVWV \
_ «I AMA A V V V 0.03O 5! “Mm—.5 A I"
E IE 7 Ex}"'-~ — I“ \‘DM—Dhfi w:
8 Zea QRQHU GNU 0.02:
E “W“ “NJ: :
E \ -6 0 -D.
IN
“HE-*8 I I | . l . I . I .
0 30 60 90 120 150
Exéosure time / h
Fig. 10
US. Patent Sep. 8, 2009 Sheet 11 of 25 US 7,585,356 B2
1E4; _I ‘ | ' ' ' I I I . I
Exposed to 15m0|% HZO+Ar at 600 °C
‘— 'Cl
“as _
Q.
TU)
IE gluhAE/Si=0.03
E 1% AWAHAaAhAWA'A"A”A""hmfimA~A~AmA—A~A—APA~AE 137 w
‘x _ [hub
:1 CI'—-
g D \ZIUW—mxmm AE/SIMO
E5*a—D~B—B~mmm~m—a-a~t3
a;Q.IN
1E~8|.Iv'-""'0 100 200 300 400 500
Exposure time / h
Fig. 11
US. Patent Sep. 8, 2009 Sheet 12 of 25 US 7,585,356 B2
1E“6 : = | . l ' I ' E ‘ I
TED
m.
J” 1E—7 - ~5
E
"('5
E
Eg 1E~8 .~ Ne -:(D 1 I
Ea:
0— Points: Expeyimental values
Curves: Caiculated value
1E-9
I I I | - l n I I I
400 500 600 700 800 900
Temperature / K
Fig. 12
US. Patent Sep. 8, 2009 Sheet 13 of 25 US 7,585,356 B2
5E”? I ' E ' I ' I ‘ I
Exposure to 75 moi% H20 at 923 K
17m
a.
TU)
E 2E-7O
E
C” aC.)
C:
8 \ 430%
E 1E-7 — 9 @Me a a 9 _
E --
CD.
IN
5E_8 I . I i I c i u I
0 30 60 90 120
Exposure time / h
Fig. 13
Sheet 14 of 25 US 7,585,356 B2
U.S. Patent Sep. 8, 2009
Vent
Vent 4—,...
Oxygen [:3 4%
Trap m :1 L153 l3
Furnace Temperature
\ Controner
5: L12.
Batance Gas w Membrane
FC
Mass Ftow
H20 2-Way Controller
\ Union Valve #1 W
__= 55 0035er _;c_ L---l
- 3; 9’0 \J TIP
Oxygen
Ice Bath 8: Wate: ‘
Trap
Carrier Gas
Oxygan
& Water
Trap
Water BathDiiute Gas
Fig. 14
U.S. Patent
15-4:
an).
’7'
0‘]
H2permeance
/m0!
m”2
3"1Pa"1
1 E?
Sep. 8, 2009 Sheet 15 of 25
1543:
E ' I I i I I | i
mStandard CVD Z
........ CRT-CV1)
.,._ .,_..W._{,
-
K.
WC /H2/CO,__
I]
[3
\:
HEICH4 .
ti
Deposition time I h
Fig. 15
US 7,585,356 B2
.100
8
Seiectévity
US. Patent Sep. 8, 2009 Sheet 16 of 25 US 7,585,356 B2
1 E-4 - i ‘ ' E i i I | r 50
1_ E I i étandard CVD —
53 nnnnnnnn ORT-CVD
‘7”, H20 conc.i=0..27 moi%
6'] O“ ' ‘ 'O o 43‘
E 1E”5 :‘ '—‘ LE2 m0! cMD/Q-E
— " 1:] win-1
0 A“ El ' 10 (.3
.'
G.)
E C /0mol% . "g
g5 3fr
g 1E—6 5— 9N
g : " \EMHH—mfia 3:
g. 0" I°:A"'-'.'.;""'A 122 mol% ME
IN (127' mol%
1 E_7 I r I . I . I . I . I ' 1
0 3 6 9 12 15 18
Deposition time I h
Fig. 16
U.S. Patent
Permeance
/mol
m'2
3'1
F‘a‘1
'1”
It»
._\.
{n
01
_"-
*1“
CD
1 E-7
1 EB
6 9 12
Deposition time / h
Fig. 17
Sep. 8, 2009 Sheet 17 of 25 US 7,585,356 B2
i I ' I ‘ I ‘ i ' I ‘ I 50
"nu H2 Depasition Temp: 673 K'
“9“ CH ‘
, “5 "A" CO
5 °\M‘-I~‘
* A‘~ ”m._ x ‘ . >,
‘ N H [CO .fl\ ‘9“ fl ‘2‘“.0 2:101“;
‘A ~ ,0 d U.— ~ W». $1,ng 3“,
,0 ------ ~ ~ ,,,,,,,u “w--__ _
I \~_.0::‘ - " ‘H
1" U ‘A‘ -.
:_ a _, ~:“®“._
: of in ‘L‘h-H “9.-1.- “~a
,- ‘ "-AU
I I I I I I 1
U.S. Patent
1E~6_
*1
U‘E
r51
--.E
*2
4
H2permeance
.1m0!m
8Pa
5Ed8
Sep. 8, 2009 Sheet 18 of 25
15:?“
I ' I ' i ' l ' I ‘
Exposure to GO main/o H20 at 673 K :
Standard CVD wlo H20
, 1,22/n'ml% H20“nu-Um..... . .....
View / ‘- a a, _
ORT with 0,2? mow/o H20 i
O 20 I 40 I 60 80 100 120 140
Exposure time i h
frig;.'1€3
US 7,585,356 B2
U.S. Patent
1E—6 ,
-2
-1
-€
H2permeance
/moim
8Pa
E“
'51
1 E—8
Sep. 8, 2009 Sheet 19 of 25
i ' i I g s I r I . i
Exposure to 60 mol% H20 at 673 K
"'13 -.
"59%‘ “A __ La H " Tut-1:: 312321;“. if"; ‘..'..".:;'.':"9.3.3::::3:33 "j
~1 8% 3
«sn- mnn» an
41%
--- a ST~L~873
~- @ ST—L—Y7’3
A ST—L~673
O 20 4O 60 80.100 120 140
Exposure time I h
Fig. 19
US 7,585,356 B2
US. Patent Sep. 8, 2009 Sheet 20 of 25 US 7,585,356 B2
“IE-6”. t . .' i ‘ I ' l ' I ' '
TiOz—SiOE: Exposure to 75 mol% H20 at 923 K -
SE02: Exposure to 15 mol% H20 at 8773 K :
“‘O~.______fl
SiO / OwW-Mmo 430%
—2
-1
—1
H2permeance/motm
3Pa
N i“~4
0—D
0 '20! 40 60 80 100 120 140
Expsoure time / h
Fig. 20
US. Patent Sep. 8, 2009 Sheet 21 of 25 US 7,585,356 B2
silica
boron titania
US. Patent Sep. 8, 2009 Sheet 22 of 25 US 7,585,356 B2
siiica
Fig. 22
US. Patent Sep. 8, 2009 Sheet 23 of 25 US 7,585,356 B2
G—Membered (a
Wfi/
- é . , é
yttrla zlrcoma
yttrla zirconia
Fig. 23
US. Patent Sep. 8, 2009 Sheet 24 of 25 US 7,585,356 B2
260 I I F I I I V I ' I ' I I I I I I I I
330“ Planar Si G-Membesed Ring Mfi M ("H - 140 " Planar SI T~membered Ring MAL.“ CH
nJ A
4
h 160- — 120- ‘7‘- CO2 .
E 140,: .- 73 MMCO
E - . E 100- N2 -
2 120- ‘_ 2 wa—O
“ wu-- ‘~ 80-
2 _
§ . _ g Ne
“c” 3”“ ‘ 2 60— "MR "— H2 —m ' GJ .. ”9‘— .
C BU~.4 c \;\ HP
.9 .9 40‘ \4 "
76 40‘ " E “‘x,
2 “ ' .2 ”\
73 20" " E 20‘ 52-. \‘x '
(E _ . (U 65“: ~ Fax»:
0 — j 0 _ mafia-flNfifi“?‘_ V -——--..1 _
'20 I I I I I I I I l ' E ' I ' I ' I V I ' I ' I
—1 D 1 2 3 4 5 B 7 B —1 D ‘I 2 3 4 5 B 7 8
distance] 10 nm distance] 10 nm
EDD ’ I ' I I ' I I I I I ' I I - I - I ' I ' I ' I I
EBB- Planar SisAIES‘Mambemd Réng min-m CH4 - 140" Pianar SiEAIJ‘membered Rang —A—— CH“ "
‘53" - 12m —V— COz -
'73 :40- - '"5 "4—CG
E E 100— N -
3 :20" 2 ?
2;, mo- _ g; 60“ ~
9’ 9g 80- : g 50..
a: (21
E 56" ' I:
2 - ,g 40— _
*5 4r}- _ *6
2 - .213 20» - 5 20w -
m . cs
0:t 6..
"
hzu................ ............,.,.
~1 0 1 2 3 4 5 6 7 8 -‘l D % 2 3 4 5 6 7 8
distance I10 nm distance! 10 nm
200 I I I l I I I I I I u I r 5 . I . l u i i I r ‘ r I q
gag- Planar Sig“l SvMembered Ring ‘_&'__ CH _ 140— P§anar SiuTi1 7—membered Ring W45.” CH '“
. 4 4 -
_ éBG— “V— C02 - 120.. ‘4‘" CO? _
'~—- —<I— CO 7- "4—4300 “10* - O
E - E 100- N2 -
:2 $20: - 2 gym «rs—~02 -
3 mn- — E, 30" ‘\ Ne -
g} . E’ .
a) Bflu - an 60- \ ”I“ H2s: E ..
w ' a: X —a— He
5 59' “ c ‘\9 - .9 40— m —
a 43- — a \\
'3 20' ”E 20 '\o .. _ .. Ax -
4: .8 Ekx \
D" . (3- WWIIWu-w—é -
“'20 "' I I I I a I I I I I I I ' I ‘ I ' I a
-1 U 1 2 3 4 5 6 "f B -1 0 I 2 3 4 5 5 T 8
distance I 10 nm distance I 10 nm
US. Patent Sep. 8, 2009 Sheet 25 of 25 US 7,585,356 B2
290 . I I I . I I I I I . I I I . I I l ' I I I I I |
130; PlanarSisB,S-Membered Ring ~-=s«~CH _ 140‘ Planar Sisal 7-membered Ring __HI___ CH ‘_ 4 _ 4
mm - 312m _
1. - Ec) “IE 140
2190—~
3 120— - % I
2, 100* - a 50' "
9’ ~ 5
2 50- T C so- -
m 50- _.g .
a a 40— ~E 40: - 3.33
'8 20- - 2°" -(U - .
0-: '- g- -
. u i Q g 3‘, g I 140— (5 I 2 3 I 5 I} i _z
150.. i’lanarSIS ,S-mambarec Ring __5,. CH _ PlanarSiuYIT—mgmbered I???) ;
disianceHO nm “ -.- 3313065 9F“ “W“ CH4 ‘
a 160— WVWCO, I 3120— “VWCOZ ..
EMD- —<a—CO — 5100 "+430 'N x " N ..
”'3 2 "x 2x mm memo - >. ‘x 2 . 9 80— "9“02 ..
3100- m’“ N8 ~ 0: -Ne .
a“) “‘“H ‘ 5 so I H- 2 .. “‘ — _' "’g 80 ‘ E} ‘ 2
E: 50“ _,\ jg 40— \W; m
g 40- \. — Q ~ "xi
U 20 “NM 3”" “wk 'm - ~ - . ‘\~ .. .
0_ H _ 0- 9“me .I
' I ' I I I I I ' I—r I I I I ' I ‘ i l I
«1 0 2 3 4 5 s 7 a 4 0 1 2 3 4 5 a 7 a
distance! i0 nm distance! 10 nm
200 I I I ' I I l ' I ' I ' I r T ' I I I I I ' I
130'“ PianarSiSZr‘s-membered Ring —‘?‘_ CHd " MD FtanarSSIZIIY—membered Ring *5“ OH:
H 160— —Lt2n~ “V—COZ -
I»...0 «41— CO
0 14a— - E
E 21430- N2 -
x 120- - ‘ —a——O
E; >~ ao- . 2 _
m 100- * g a Ne
E 80 F g 60“ mumH _
”9 c: . .
,5 5°" “3% 40~ ”‘ -"I6 " > .
2 40‘ “g 20
a 20- - m _ I
o— -— 0- -
I ' I ' I l l I I K
.1 8 4 0 1 2 3 4 5 5 7 B
distance I 10 nm disiance! 10 nm
US 7,585,356 B2
1
HYDROTHERMALLY—STABLE
SILICA-BASED COMPOSITE MEMBRANES
FOR HYDROGEN SEPARATION
CROSS-REFERENCE TO RELATED
APPLICATIONS
Not applicable.
STATEMENT REGARDING FEDERALLY
SPONSORED RESEARCH OR DEVELOPMENT
Not applicable.
TECHNICAL FIELD OF THE INVENTION
The present invention relates to composite membranes
applicable for the separation of gases. The composite mem-
brane comprises an overcoat of a silica-based composite
placed on optional porous substrates deposited on a porous
support. The invention also includes methods for preparing
such composite membranes. The composite membrane can
be used for the selective separation of hydrogen from other
gases, and is particularly advantageous for use in humid envi-
ronments.
BACKGROUND OF THE INVENTION
Membranes may be defined as thin, solid materials that
permit the selective transport ofcertain chemical species over
others.
Silica membranes prepared by chemical vapor deposition
(CVD) or sol-gel methods on mesoporous supports are effec-
tive for selective hydrogen permeation [T. Okubo and H.
Inoue, J. Membr. Sci., 42 (1989) 109; G. R. Gavalas, et al.,
Chem. Eng. Sci., 44 (1989) 1829; S. Yan et al., Ind. Eng.
Chem. Res., 33 (1994) 2096]. However, it is known that
hydrogen-selective silica materials are not hydrothermally
stable. Most researchers have reported a loss of permeability
of silica membranes (as much as 95% or greater in the first 12
h) on exposure to moisture at high temperature Sea et al.,
found that the hydrogen permeance of a silica membrane
deposited on mesoporous y-alumina substrates was decreased
by 90% from 3.5><10'7 to 4.0><10'8 mol m‘2 s'1 Pa‘1 after
exposure to 50 mol % water vapor at 400° C. for 100 h [Sea et
al. Ind. Eng. Chem. Res. 37 (1998) 2502]. Wu et al. reported
a decrease of 62% and 70% in the permeances of He and N2
for a CVD-deposited silica membrane treated at 600° C.
under a N2 flow containing 20 mol % water vapor [Wu et al.,
J. Membr Sci., 96 (1994) 257]. This is because the porous
silica (SiOZ) easily undergoes densification upon exposure to
water vapor at elevated temperatures. The densification
involves the formation of SiiOiSi bonds from silanol
groups (SigOH) catalyzed by water, leading to the shrinkage
of pores (Her, The Chemistry of Silica, Wiley, New York,
1 979).
Much effort has been expended on the improvement of the
stability ofsilica membranes. One approach is to make hydro-
phobic silica membranes prepared by the incorporation of
methyl groups in the silica microstructure [de Vos et al., J.
Membr Sci., 158 (1999) 277].], [Y. S. Lin, I. Kumakiri, B. N.
Nair, H. Alsyouri, “Microporous Inorganic Membranes”,
Separ. Purif, Methods, 2002, 31, 229-379].
On the other hand, composite membranes prepared by
sol-gel methods composed of silica with other inorganic
oxides such as alumina (A1203), titania (TiOz) and zirconia
(ZrOz) have been reported as better alternatives to silica
10
20
25
35
40
45
55
2
membranes for use under humid atmospheres at high tem-
perature. Fotou et al. [Fotou et al., J. Mater Sci., 30 (1995)
2803] introduced these oxides and MgO into the membranes
by doping the starting silica sol with controlled amounts of
the corresponding nitrate salts. They found that the mean pore
size did not change much from 0.6 nm to 0.7 nm, but the
hydrothermal stability was improved after doping with 30%
alumina.A heat treatment in 50 mol % steam/air at 6000 C. for
30 h resulted in 63.6% reduction in the surface area and a loss
of86.5% micropore volume for the unsupported3% alumina-
doped silica membrane, compared to 84.6% and 94.5%,
respectively for a pure silica membrane. They also reported
that 6% alumina-doped and magnesia-doped silica mem-
branes were not improved, since the surface area was sub-
stantially reduced compared with the pure silica. The mem-
branes prepared by Fotou et al. differ from our membranes
because they consist of layers of sol particles without a con-
tinuous toplayer deposited by CVD. As such they have spaces
in between the particles that give rise to poor selectivity. The
authors do not report permeance or selectivity, however, these
properties have been measured for similar membranes. The
permeation properties of such membranes prepared by the
sol-gel deposition of SiOZ-10 mol % A1203 and SiOZ-10 mol
% TiO2 compositions on a gamma-alumina support were
reported by Hekkink et al., [Hekkink et al., Key Eng. Mater.,
61&62 (1991) 375]. The H2 permeances at 298 K were
7><10'7, 2.2><10'7 and 6><10'8 mol m"2 s"1 Pa"1 for pure SiOz,
SiOZ-10 mol % TiO2 and SiOZ-10 mol % A1203 membranes,
respectively. The SiOziAle3 derived membrane had per-
meance for H2 of2.5><10'7 mol m‘2 s'1 Pa‘1 at 301 K and the
SiOziTiO2 derived membrane hadpermeance for H2 of6.7><
10'7 mol m"2 s"1 Pa"l at 473 K. These permeances are high,
but the selectivities over CO were only 3 and 9. This is
indicative of the presence of channels that permit passage of
all gases, and is typical for membranes prepared by the sol-gel
method. Selectivity can be increased, but only at the cost of
reducing permeance, as discussed below.
As another example of work on sol-gel membranes,
Yoshida et al. investigated the hydrothermal stability of sol-
gel derived silica-zirconia membranes with a content of Zir-
conia of10-50 mol % [Yoshida et al., J. Chem. Eng. Japan, 34
(2001) 523]. After a 20 h-exposure to a high temperature of
773 K and steam at levels of 13-33 mol %, a SiOZ-10 mol %
ZrO2 membrane still suffered a decrease of H2 permeance of
70% to 8.9><10'8 mol m‘2 s'1 Pa'l, but with an increased
selectivity ofH2 to N2 of 190, while the SiOZ-50 mol % ZrO2
membrane did not show any change in the H2 permeance but
a constant HZ/N2 selectivity of 4.0. It is well known in the
membrane field that selectivity can be increased with a drop
in permeance, and this represents a typical example of this
phenomenon.
The composite membranes prepared by the chemical vapor
deposition method generally have a better selectivity but a
lower permeance in comparison to those obtained with the
sol-gel procedure Nam et al., made SiOziTiO2 membranes
at 873 K on porous Vycor glass with a mean pore diameter of
4 nm by hydrolysis of tetraisopropyl titanate (TIPT) and
tetraethyl orthosilicate (TEOS) at atmospheric pressure [Nam
et al., Korean Membr. J ., 3 (2001) 69]. Using molar ratios of
TIPT/TEOS in the range of 0.1-7, composite membranes
were obtained that showed high selectivities of around 500
but with low permeance of2><10'8 mol m"2 s"1 Pa"l at 873 K.
This again is an example of the tradeoff between selectivity
and permeance that gives rise to high selectivity at the cost of
low permeance.
To obtain ceramic membranes with both high selectivity
and permeance, some researchers have used mesoporous or
US 7,585,356 B2
3
macroporous supports with pore diameters larger than 50 nm
to decrease the resistance of the supports. By using an inter-
mediate mesoporous gamma-alumina layer, Yan et al. and
placing the silica in the pores of the support by CVD they
obtained a H2 permeance of 1.8><10'7 mol m‘2 s'1 Pa'1 but a
selectivity ofH2 to N2 ofonly 26 at 873 K. [Yan et al. Ind. Eng.
Chem. Res., 33 (1994) 2096]. The poor selectivity was due to
the presence of defects due to the use ofan intermediate layer
of large particle size, which left large openings between the
particles. Oyama et al., made membranes with the silica layer
on the outer surface of an alumina substrate and obtained
permeability of2.2><10'7 mol m‘2 s'1 Pa'1 and a selectivity of
H2 to CO of370 at 873 K [S. T. Oyama, L. Zhang, D, Lee, D.
S, Jack, “Hydrogen Selective Silica-Based Membranes” US.
Pat. No. 6,854,602 B2, Feb. 15, 2005]. Recently, we have
successfully prepared a gamma-alumina multilayer with a
graded structure by sequentially placing boehmite sols of
gradually decreasing particle sizes on a macroporous alumina
support Oyama et al. [8. T. Oyama,Y Cu, D. Lee, US. patent
application Ser. No. 10/775,288, Feb. 10, 2004]. The multi-
layer graded structure had a thickness of tens of nanometers
and was substantially defect-free. After deposition of a thin
silica layer by the CVD technique method described in a
patent [Hydrogen-Selective Silica Based Membrane, S. T.
Oyama, A, Prabhu, US. Pat. No. 6,527,833, Mar. 4, 2003],
the resulting silica-on-alumina membranes had excellent per-
meability of 3.0><10'7 mol m"2 s"1 Pa"1 and good selectivity
for hydrogen over CH4, CO and CO2 of over 500 at 873 K.
SUMMARY OF THE INVENTION
This invention relates to composite silica-based mem-
branes for the separation of hydrogen and other gases which
are stable to high temperature in moist or humid environ-
ments The invention comprises mixed-element membranes
composed of silica and another element or elements depos-
ited on an optional porous substrate by chemical vapor depo-
sition of gaseous precursors. The porous alumina substrate is
in turn placed on top of a support, which may be any porous
material. The purpose of the substrate is to provide a more
uniform surface on which to place the mixed-element com-
position.
According to preferred embodiments of this invention,
SiOziAle3 and SiOziTiO2 composite membranes are
prepared on graded and nongraded mesoporous gamma-alu-
mina intermediate layers supported on macroporous alpha-
alumina tubes by employing a dual-element CVD technique
at high temperature. In some embodiments, the dual-element
CVD technique at high temperature utilizes opposing reac-
tant flows.
BRIEF DESCRIPTION OF THE DRAWINGS
For a better understanding of the present embodiments,
reference is made to the accompanying drawings, in which:
FIG. 1 is a plot showing particle size distribution of boeh-
mite sols labeled BS40, BS200 and BS630;
FIG. 2 is a schematic diagram of a mechanical motor-
driven dip-coating machine;
FIG. 3 is a plot showing pore size distributions ofy-alumina
membranes prepared from boehmite sols BS40. BS200 and
BS630;
FIG. 4 is a micrograph showing a cross-sectional image of
a gamma-alumina multilayer substrate formed on a
macroporous alumina tube;
FIG. 5 is a schematic diagram of a dual-element CVD
apparatus used in the deposition of the silica-alumina com-
posite layer;
25
30
50
55
4
FIG. 6 is a plot showing permeation properties at 873 K of
a silica-alumina composite membrane prepared with a ratio
of ATSB/TEOS:0.03 and a TEOS concentration of 19.4><
10'3 mol m"3 ;
FIG. 7 is a plot showing permeation properties at 873 K of
a silica-alumina composite membrane prepared with a ratio
of ATSB/TEOS:0.02 and a TEOS concentration of 19.4><
10'3 mol m"3 ;
FIG. 8 is a plot showing H2/CH4 selectivity and H2 per-
meance at 873 K ofthe composite membranes prepared with
different molar ratios ofATSB/TEOS;
FIG. 9 is a plot showing permeation properties at 873 K of
a silica-alumina composite membrane prepared with a ratio
ofATSB/TEOS:0.02 and a TEOS concentration of 13><10'3
mmol m‘3;
FIG. 10 is a plot showing changes in H2 permeance ofpure
silica and composite silica-alumina membranes prepared
using different molar ratios ofATSB/TEOS;
FIG. 11 is a plot comparing the long-term hydrothermal
stability ofthe pure silica membrane and the composite mem-
brane SA-IV obtained using CVD Condition IV with a molar
ratio ofATSB/TEOS of 0.03;
FIG. 12 is a plot showing the permeability of He, H2, and
Ne through a silica-alumina membrane deposited on a
gamma-alumina multilayer substrate by using a dual-element
CVD technique using Condition VI;
FIG. 13 is a plot showing changes in H2 permeance of
composite silica-titania membranes prepared at 923 K using
molar ratios ofATSB/TEOS of 0.05
FIG. 14 is a schematic illustration ofa multi-element CVD
apparatus using opposing reactants technique (ORT) for use
in the deposition of titania-silica layers;
FIG. 15 is a plot showing changes in H2 permeance and
HZ/CH4 selectivity with the deposition time of composite
membranes prepared at 873 K by using a standard CVD and
ORT-assisted CVD;
FIG. 16 is a plot showing changes in H2 permeance and
HZ/CH4 selectivity with the deposition time ofthree compos-
ite membranes prepared at 773 K by using a standard CVD
and ORT-assisted CVD with the use of different water con-
centration;
FIG. 17 is a plot showing changes of permeances and
selectivities with the deposition time of a composite mem-
brane prepared at 673 K by using ORT-assisted CVD;
FIG. 18 is a plot showing the effect of water vapor on
changes in H2 permeance during exposure to 60 mmol % HZO
at 673 K for three composite membranes prepared at 773 K by
using a standard CVD and an ORT-CVD;
FIG. 19 is a plot showing changes in H2 permeance during
the exposure to 60 mol % HZO at 673 K for three composite
membranes prepared at 673-873 K by using ORT-assisted
CVD;
FIG. 20 is a plot showing changes in H2 permeance during
the exposure to 75 mol % HZO at 923 K for a composite
membrane prepared at 923 K by a standard CVD with the use
ofmolar ratio of TIP/TEOS of 0.065;
FIGS. 21-23 are illustrations of optimized structures of
6-membered silica rings containing Al, B, and Ti, 7-mem-
bered silica rings containing Al, B, and Ti, and 6- and 7-mem-
bered silica rings containing Y and Zr, respectively; and
FIGS. 24-25 are a series of plots showing activation ener-
gies for permeation ofvarious gaseous species through silica
6-membered and 7-membered rings, silica-alumina 6-mem-
bered and 7-membered rings, silica-titania 6-membered and
7-membered rings.
US 7,585,356 B2
5
DETAILED DESCRIPTION OF THE PREFERRED
EMBODIMENTS
The thin and defect-free composite membranes of the
present invention are formed on an optional intermediate
substrate that can be a single layer or a multilayer with a
gradually decreasing pore size structure placed on a
macroporous support. The substrate is prepared by coating a
mesoporous support with particles so as to make its external
pore structure (in the surface region) more uniform. The
particles can be in the nanometer to micrometer size range. A
single layer of particles may be used of varying or constant
size. Alternatively, multiple layers of particles of different
size may be used to achieve the end of obtaining a uniform
surface layer. Preferably, the layers will be composed of
particles of increasingly small size. The particles are formed
in any manner, but will be denoted here as sols. A sol is
defined here as collection of suspended particles. The inven-
tion uses a single dilute sol dipping solution or a series of sol
dipping solutions containing sols of different particle sizes.
The dipping solutions are used to coat a support. Each coating
step is followed by a calcination step.
There is no restriction on the support except that it be
porous. Although not restricted to these, the porous support
comprises a material selected from the group consisting of
alumina, silica, titania, magnesia, zirconia, zeolites, carbon,
phosphorus, gallium, germanium, yttria, niobia, lanthana,
stainless steel and combinations thereof. These are well-
known materials that can be made in porous form.
The use of an intermediate layer results in a uniform,
microporous substrate on which to deposit the topmost layer.
This allows the present membranes to be thin and defect-free
and thus have both high selectivity and permeance. Ourmem-
branes also make use ofthermal decomposition in inert gas of
the silica, alumina, and titania precursors, rather than oxida-
tive degradation. As is shown in EXAMPLE 11 below, the use
of oxygen results in membranes having poor permeance.
In cases where the support is itself uniform and
microporous, it should be evident to those skilled in the art
that the intermediate substrate would be optional.
The sols can be ofany composition as described above. An
example will be given here for boehmite (AlOOH) sols, Boe-
hmite sols with different particle sizes were prepared by
carefully controlling the hydrolysis of aluminum alkoxides
and the subsequent acid peptization ofthe resulting boehmite
precipitate. The general procedure for preparing boehmite
sols was as follows. A quantity of 0.2 mol of aluminum
isopropoxide (Aldrich, 98+) was added to 300 ml of distilled
water at room temperature. The mixture was quickly heated to
353 K within 0.5 h with high speed stirring and was main-
tained at this temperature for 3-24 h for the hydrolysis of the
isopropoxide and the formation of a boehmite (AlOOH) pre-
cipitate. The precipitate was then heated to 365 K and was
peptized using a quantity of acetic acid (GR, 99.7%) with a
molar ratio of H+/Alkoxide in the range of 0.04-0.15 Pepti-
zation refers to the breakup oflarge oxide precipitate particles
by acid treatment.
The solution was refluxed at 365 K for 20 h to get a clear or
slightly translucent sol. The concentration of the resulting
boehmite sols was calculated from the volume of the liquid
and the known quantity of isopropoxide used. A dynamic
light scattering analyzer (Horiba Model LB-500) was used to
measure the particle size of the boehmite sols. These sols
remained stable for more than 3 months. Three boehmite sols
with median particle sizes of40, 200 and 630 nm were used in
the present invention to prepare the gamma-alumina multi-
layer substrate. FIG. 1 shows the particle size distributions of
these three boehmite sols. The sols labeled BS40, BS200 and
BS630 have median particle diameters of40 nm, 200 nm and
630 nm, respectively.
25
35
45
55
6
A thin and defect-free gamma-alumina multilayer
employed in the present invention was prepared on a
macroporous alpha-alumina support by the dipping-calcining
method described in patent application number PCT/USOS/
03552. The dip-coating of the support was carried out with a
series of dilute dipping solutions containing boehmite sols of
different particle sizes., A commercial alumina membrane
tube (PALL Corporation, Membralox® TI-70-25Z Mem-
brane Tube, I.D.:7 mm, O.D.:10 mm) with a nominal pore
size of 100 nm was used as the support. The preparation
involved several steps. First, the alumina tube was cut to a
length of 3-4 cm with a diamond saw and was connected to
non-porous alumina tubes at both ends with ceramic joints.
The ceramic joints were made with a glaze (Duncan IN 1001)
filed at 1153 K for 0.5 h. Second, dilute dipping solutions
were prepared by mixing the boehmite sols with a polyvinyl
alcohol (PVA, M.W.:72,000) solution and diluting with dis-
tilled water to obtain a 0.15 M concentration of the sol and a
0.35 wt. % concentration of the PVA. Third, the alumina
support was dipped into the dipping solution and was with-
drawn after 10 seconds at a late of 0.01 m s"1 to ensure
uniform and reproducible coatings using a motor-driven dip-
coating machine. The dip-coating machine was built in—house
and used a stepping motor drive (FIG. 2) Fourth, the dip-
coated alumina was dried in ambient air for 24 h, and thenwas
heated to 923 K in air at a rate of 1 Kmin'l and calcined at 923
K for 2 h. The dipping-calcining process was successively
carried out five times using different dipping solutions con-
taining boehmite sols in the order of decreasing sol particle
size: 630, 630, 200, 40 and 40 nm. (Two of the sols were
applied twice).
The microstructures of the intermediate gamma-alumina
membranes were characterized by nitrogen physisorption
conducted in a volumetric unit (Micromeritics ASAP 2000).
The samples were prepared using the same procedure and
parameters as the supported membranes, First, a boehmite sol
was cast on a glass Petri dish and dried at ambient temperature
in air. The dried gel flakes were recovered from the bottom of
the Petri dish, and were then heated to 923 K in air at a rate of
1 K min‘1 and maintained at this temperature for 2 h. The
Barrett, Joyner and Halenda (BJH) method was used to deter-
mine the pore size distribution using the desorption isotherm.
FIG. 3 illustrates the pore size distributions of the gamma-
alumina substrates prepared from the boehmite sols contain-
ing particles of size of 630, 200, and 40 nm. These substrates
had a sharp pore size distribution. Table 1 lists the microstruc-
ture parameters ofthese three substrates. It was found that the
larger the particle of sols, the larger the pore size and porosity
of the resulting membranes as discussed in connection with
FIG. 1.
TABLE 1
Microstructure Parameters of Gamma-Alumina
Membrane Calcined at 923 K for 2 h
Gamma-alumina membrane calcined at 923 K for 2 h
Sol Pore Pore Average
particle size volume surface area pore size Porosity*
(Hm) (0m3 g’l) (m3 g’l) (Hm) (%)
630 0.4731 370.4 5.11 63.6
200 0.4321 378.7 4.56 61.5
40 0.3622 388.6 3.73 57.3
pgammafllumma = 3.7 g cm’3 (R. S. A. de Lange et al., J. Membr. Sci., 99
(1995) 57)
The cross-sectional microstructure of the gamma-alumina
multilayer substrate was characterizedusing a Field Emission
Scanning Electron Microscope (FESEM, Leo 1550). The
samples were coated by sputtering with a layer ofgold before
US 7,585,356 B2
7
measurement with the electron microscope. The thicknesses
ofthe y-alumina substrate were obtained from cross-sectional
photos with high resolution. FIG. 4 shows a cross-sectional
image of the 5-layer graded y-alumina membrane substrate.
The composite membranes of the present invention were
prepared using the previously described gamma-alumina
multilayer substrates by the deposition of a thin multicompo-
nent layer by the chemical vapor deposition (CVD) method.
This process places a multicomponent layer on the surface of
the substrate by thermal decomposition at high temperature.
One ofthe components is silica in the form oftetraethylortho-
silicate (TEOS). The silica source is not restricted to TEOS. It
can be any volatile compound of silica. Although not limited
to these, it can be tetramethylorthosilicate (TMOS), ethyltri-
ethoxysilane, silane, chlorosilane, and combinations thereof.
The other components are not restricted. Examples are pre-
sented for alumina and titania and theoretical calculations are
presented for other elements. The aluminum compound used
was aluminium-tri-sec-butoxide (ATSB) but the alumina
source is not restricted to ATSB. Although not limited to
these, it can be aluminum tributoxide, aluminum tri-tert-bu-
toxide, aluminum triethoxide, aluminum chloride, and com-
binations thereof. The titanium compound used was titanium
isopropoxide (TIP) but the titania source is not restricted to
TIP. Although not limited to these, it can be other titanium
alkoxides, titanium alkyls, titanium chloride, and combina-
tions thereof. It should be clear to those skilled in the art that
alumina and titania are not the only materials that can be
combined with silica. Other oxides that can be employed are
those of B, Al, P, Ga, Ge, As, In, Sn, Sb, Sc, Ti, V,Y, Zr, Nb,
La, Hf, and Ta, that can form bonds with silica, Similarly, it
should be clear to those skilled in the art that the combinations
are not restricted to binary combinations, and that multiple
component combinations are possible.
The CVD setup is shown in FIG. 5, and the CVD process
parameters are listed in Table 2.A specific example is cited for
a silica-alumina composite. The process parameters are not
restricted to those listed. It should be clear to those skilled in
the art that the parameters can vary over a broad range. The
range can also be affected by the scale of the synthesis, the
speed of the desired transformation, the geometry of the
components, the particular apparatus used, and the expense of
the reagents, and other considerations. The CVD conditions
themselves are also not restricted to 873 K and atmospheric
pressure. It should be clear to those skilled in the art that the
10
15
20
25
30
35
40
8
temperature can range from 298 K to 1273 K and the pressure
from subatmospheric to 30 atm.
The support coveted with the gamma-alumina layers was
installed concentrically inside a piece of quartz tubing of 14
mm inside diameter using machined Swagelok fittings with
Teflon ferrules. After placing the assembly in an electrical
furnace and heating it to 873 K at a heating rate of 1 K min—1,
an argon gas flow was introduced on the outer shell side and
a dilute argon gas Dow was passed on the inner tube side.
After 30 min. a TEOS carrier gas flow was passed through a
bubbler filled with TEOS at 296 K and a separate ATSB
carrier gas flow was passed through a bubbler filled with
ATSB at a higher temperature in the range of357-369 K. The
two carrier gases were then premixed with the dilute Ar flow
before introduction to the inside of the support tube. The
molar ratio of ATSB to TEOS was adjusted by carefully
controlling the flow rates ofthe carrier gases and the tempera-
ture ofATSB. The deposition time was varied from 3 to 6 h.
After the CVDprocess was finished, the assembly was purged
with the balance and dilute gas flows for 30 min. The gas
permeation measurement was generally conducted at 873 K
on H2, CH4, CO and CO2 by admitting the pure gases at a
certain pressure (higher than atmospheric pressure) into the
inner tube side, one end ofwhich was closed, and measuring
the quantity of gas flowing from the outer tube. The selectiv-
ity was calculated as the ratio ofthe perrneances ofH2 to CH4,
CO and C02. Permeation ofHe, H2, and Ne was measured in
a similar manner at different temperatures.
The hydrothermal stability test was carried out under anAr
flow containing 16 mol % or 57 mol % water vapor at 873 K
up to 520 h. First, anAr flow at 15 mol s"l (flow rates in umol
s"1 can be converted to cm3 (NTP) min"1 by multiplication by
1.5) was passed through a heated bubbler containing distilled
water and was then introduced on the inner membrane tube
side to directly contact the fresh composite membranes, while
another Ar flow also at 15 umol s"1 was maintained on the
outer shell side. The H2, CH4 and CO2 permeation rates were
measured periodically during the hydrothermal stability test
to monitor the changes in the perrneance and selectivity. To
make the measurements, water vapor was shut off for about
20 min to dry the membranes under a dry Ar flow. The wet Ar
flow was resumed immediately after the perrneance measure-
ments.
TABLE 2
CVD Process Parameters for the Preparation of Silica-Alumina Membranes
(CVD temperature is always 873 K)
Cond. Cond. Cond. Cond. Cond. Cond. Cond.
I II III IV V VI VII
TEOS Bath Temp. (K) 296 296 296 296 296 296 296
ATSB Bath Temp. (K) 365 369 367 363 358 357 357
TEOS Carrier Gas 2.9 3.6 3.7 3.7 3.7 3.3 3.3
(Hmol 571)
ATSB Carrier Gas 10.3 5.7 4.1 4.0 4.2 4.1 4.1
(mol 54)
Dilute Gas 21.4 7.7 9.2 9.2 9.2 15.5 25.4
(Hmol 571)
Balance Gas 34.4 17.0 17.0 16.9 17.1 22.9 32.8
(mol 54)
TEOS Concen. 7.5 19.2 19.4 19.4 19.4 13.0 9.1
><103 (mol m’3)
ATSB Concen. 0.83 1.25 0.77 0.57 0.40 0.27 0.19
><103 (mol m’3)
ATSB/TEOS 0.11 0.065 0.04 0.03 0.02 0.02 0.02
(molar)
US 7,585,356 B2
9
EXAMPLE 1
This example describes the synthesis ofboehmite sols that
are used for the preparation ofthe gamma-alumina multilayer
substrates. A boehmite sol was prepared by adding 0.2 mol of
aluminum isopropoxide (Aldrich, 98+) to 300 ml of distilled
water at room temperature. The mixture was stirred at high
speed and heated to 353 K. The alkoxide was hydrolyzed at
this temperature for 3 or 24 h, and then the temperature ofthe
mixture was raised to 365 K after which the flask was opened
for 1.5 h to allow volatilization of the alcohol. The flask was
then closed again and the solution was stirred at 365 K for 1
h with refluxing. Then, a quantity of acetic acid (GR, 99.7%)
was added to the solution to give a 0.15, 0.07 or 0.04 ofmolar
ratio of H+/Alkoxide. After peptization at 365 K with reflux-
ing for 20 h, a clear or slightly translucent stable sol solution
was obtained, Three boehmite sols with a median particle size
of 40, 200 and 630 nm denoted as B540, BS200 and BS630,
respectively were obtained by carefully controlling the syn-
thesis parameters, as listed in Table 3. The results in Table 3
demonstrate that long hydrolysis time and low acid concen-
tration produce larger sol particles. Due to the relatively long
time of peptization, these colloid sols have a narrow particle
size distribution, as shown in FIG. 1. They were found to be
stable for more than three months. They were used for the
preparation ofthe intermediate layers ofthe gamma-alumina
membrane substrate as will be described,
TABLE 3
Synthesis Parameters of Boehmite
Sols with Different Particle Size
Molar ratio ofHydrolysis time Average median
Sample (h) H+/Alkoxide particle Size (mm)
BS40 3 0.15 40
BS200 24 0.07 200
BS630 24 0.04 630
EXAMPLE 2
This example describes the preparation of dipping solu-
tions which are used in the dipping-calcining procedure for
placing alumina layers on top of a porous support. The dip-
ping solutions are diluted combinations of the sol solutions
described in Example 1 mixed with a binder, polyvinyl alco-
hol (PVA). Three dipping solutions with sol concentrations of
0.15 M were prepared. The dipping solution made using the
boehmite sol BS40 with median particle size of 40 nm
described in Example 1 was designated as DS40. 400 ml of
the dipping solution DS40 were prepared as follows. First, 3.5
g ofPVA (Fluka, M.W.:72,000) and 5 ml of 1 M HNO3 were
added to 95 ml of boiling water with vigorous stirring and
refluxing. After 4 h, a clear solution with a PVA weight
concentration of 3.5% was obtained. Then, 77 ml of 0.80 M
boehmite sol BS40 were vigorously mixed with 283 ml of
distilled water and 40 ml of the 3.5 wt % PVA solution were
added and refluxed at 323 K for 2 h. The final concentrations
of PVA and boehmite sol were 0.35 wt % and 0.15 M. The
solution was cooled to room temperature at quiescent condi-
tions for 1 h and was set aside for the preparation of the
gamma-alumina membranes.
Dipping solutions DS630 and DS200 were obtained by the
same procedure using the boehmite sols BS630 and BS200
with median particle sizes of 630 and 200 nm described in
Example 1.
5
10
15
20
25
30
35
40
45
50
55
60
65
10
EXAMPLE 3
This example describes the preparation of intermediate
gamma-alumina multilayer substrates with a graded structure
by the deposition of five gamma-alumina layers oil top of a
porous support, where the layers are formed from dipping
solutions with increasingly small sol particles. The support
used was a commercial alumina membrane tube with a nomi-
nal pore size of 100 nm. It was used in sections of 3-4 cm
length connected with non-porous alumina tubing at both
ends by ceramic joints. A dip-coating method was employed
to coat the alumina supports with the sol and binder materials.
First, the dipping solution D8 630 containing the boehmite
sol with median particle size of630 nm described in Example
2 was used. The support was dipped at a speed of 0.01 m s"1
in the dipping solution DS630, was held for 10 seconds, and
was withdrawn at the same speed. Use was made of the
mechanical dip-coating machine shown in FIG. 2. The sol-
coated tube was dried in ambient air for 24 h, heated to 923 K
at a heating rate of 1 K min"1 and calcined for 2 h. Then, the
dipping-calcining procedure was repeated using the same
dipping solution DS630, followed by application of the dip-
ping-calcining procedure with solutions DS220, D840 and
D840. As described in Example 2, the solutions DS630, D
S200 and D840 contained the sots with median particle sizes
of630, 200 and 40 nm, respectively. The pore size distribution
of every layer of this multilayer substrate is given in FIG. 3.
The cross-sectional image of this multilayer substrate
obtained by high resolution field emission scanning electron
microscopy (FESEM) is shown in FIG. 4. The figure shows
three regions of y-alumina (corresponding to the layers
formed with three solutions of different particle size) with a
clear boundary formed on the top of the macroporous ot-alu-
mina support. The textures of the y-alumina layers are differ-
ent and the top layers ofthe figure have finer structure than the
bottom layers ofthe figure. The total thickness ofthis 5-layer
section is only around 1200 nm (1.2 pm). Also, FIG. 4 shows
that there is a good connection between the y-alumina multi-
layer and the ot-alumina support and that no infiltration of sol
particles (or pore clogging) occurred in the pores of the sup-
port during the dip-coating process.
EXAMPLE 4
This example describes the preparation of three different
silica-alumina composite membranes by the mixed chemical
vapor deposition (CVD) technique of this invention. The
composite membranes were deposited on the same alumina
multilayer substrate described in EXAMPLE 3, but different
CVD process parameters denoted Condition I, Condition II
and Condition III, and listed in Table 2 were employed sepa-
rately. The composite membrane SA-I was obtained by using
Condition I. The silica source was TEOS (tetraethylorthosili-
cate) vapor and was introduced using a bubbler at 296 K with
Ar as a carrier gas at a flow rate of2.9 umol s'l. The alumina
source was ATSB (aluminum-tri-sec-butoxide) vapor, and it
was similarly introduced by means ofa bubbler at 365 K using
Ar as a carrier gas at a flow rate of 10.3 mmol s'l. The TEOS
and ATSB flows were combined and passed on the tube (in-
ner) side ofthe tubular membrane, while a flow ofAr at a rate
of 34.4 umol s'1 was maintained on the shell (outer) side of
the reactor assemblyA dilution gas flow at a rate of21 .4 umol
s'1 was also used so that the TEOS concentration was 7.5><
10'3 mol m"3 and ATSB concentration was 0.83><10'3 mol
m'3, leading to a molar ratio of ATSB/TEOS of 0.11. The
dual-element CVD process was conducted for 3 h and 6 h
with the apparatus shown in FIG. 5. The permeation proper-
US 7,585,356 B2
11
ties at 873 K before and after CVD aye listed in Table 4. The
selectivities of H2 over CH4 and CO2 for the membrane sup-
port with the intermediate gamma-alumina multilayer are
close to the values predicated by the Knudsen diffusion
mechanism. After 3 h and 6 h of CVD, the permeance
declined slightly but the selectivities did not improve much.
The H2 permeance was high, 2.6><10'6 mol m‘2 s'1 Pa'1 at
873 K.
The same procedure was followed to deposit a layer of
silica-alumina using the CVD process Condition II and Con-
dition III. After adjusting the flow rates of the ATSB carrier
gas and the dilution gas, the TEOS concentrations were kept
at around 19><10'3 mol m"3 while the ATSB concentrations
were 1.25><10'3 and 0.77><10'3 mol m'3, giving rise to molar
ratios ofATSB/TEOS of 0.065 and 0.04 for Condition II and
Condition III, respectively. The resulting composite mem-
branes were designated composite membranes SA-II and SA-
III, respectively. The permeation properties at 873 K are also
listed in Table 4.
The composite membranes prepared with the molar ratio of
ATSB/TEOS in the range of 0.04-0.11 had relatively high
permeability but low selectivity.
TABLE 4
10
15
20
12
TABLE 5
Gas Permeation Properties of a Silica-Alumina Composite
Membrane Before and After Dual-Element CVD at 873 K
Multilayer SA-IV
Permeation properties substrate 3 h-CVD 6 h-CVD
Permeance H2 4.5 x 10*5 1.2 x 10*6 2.1 x 10*7
(mol m’2s’lPa’l) CH4 1.6 x 10*5 2.9 x 10*7 1.2 x 1059
C02 9.6x10’6 2.0)(10’7 1.2x10’9
Selectivity H2/CH4 2.8 4.2 170
H2/CO2 4.7 6.2 180
EXAMPLE 6
This example describes another silica-alumina composite
membrane deposited using CVD ConditionV on the gamma-
alumina multilayer substrate described in EXAMPLE 3. The
same TEOS concentration of 19.4><10'3 mol m"3 was used as
Gas Permeation Properties of Silica-Alumina Composite
Membranes Before and After Dual-Element CVD at 873 K
Multilayer SA-I
Permeation properties substrate 3 h-CVD 6 h-CVD
SA-II
3 h-CVD
SA-III
3 h-CVD 6 h-CVD
Permeance H2 4.5 x10’5 6.0 x 10*6 2.6 ><10’6 2.8 x 10*7 1.4 ><10’6 4.2 x 10*7
(mol - m’2s’lPa’l) CH4 1.6 x10’5 2.1)(10’6 6.3 ><10’7 1.7 ><10’8 2.3 ><10’7 3.8 ><10’8
co2 1.0 x1075 1.3 ><10’6 2.8 ><10’7 9.2 ><10’9 1.7 ><10’7 2.3 ><10’8
Selectivity H2/CH4 2.8 2.8 4.1 16 6.1 11
H2/C02 4.5 4.6 9.3 30 8.3 18
EXAMPLE 5
This example describes another silica-alumina composite
membrane deposited on the gamma-alumina multilayer sub-
strate described in EXAMPLE 3. Condition IV was used in
this case and the resulting composite membrane was desig-
nated SA-IV. Compared with the cases in EXAMPLE 4, a
lower molar ratio of ATSB/TEOS (0.03) was used in this
example. TEOS was introduced through a bubbler at 296 K
usingAr as a carrier gas at 3 .7 umol s"1 on the inner side ofthe
tubular membrane, and ATSB through a bubbler at 363 K
using Ar as a carrier gas at 4.0 umol s'l, while a flow ofAr of
16.9 umol s"1 was maintained on the shell side ofthe reactor
assembly.A dilution gas at a rate of 9.2 umol s'1 was added so
that the TEOS concentration was 19.4><10'3 mol m"3 and the
ATSB concentration was 0.57><10'3 mol m‘3, leading to a
molar ratio ofATSB/TEOS of 0.03. The dual-element CVD
process was conducted for 3 h and 6 h. FIG. 6 and Table 5
show the changes of permeation properties at 873 K before
and after CVD. The membrane had a SiiAl CVD layer
deposited on a gamma-alumina multilayer placed on top of a
macroporous alumina support, CVD preparation condition
IV was used with molar ratio of ATSB/TEOS of 0.03 and
TEOS concentration of 0.0194 mol m'3. The selectivities
increased quickly with the deposition time. After 6 h ofdepo-
sition, the selectivities of 112 over CH4 and CO2 were 170,
and 180, respectively at 873 K with the H2 permeance of
2.1><10'7 mol m'2 s'1 Pa'l, which was better than that in
Example 4.
40
45
50
55
60
in EXAMPLE 5, but a lower molar ratio ofATSB/TEOS of
0.02 was employed. TEOS was introduced through a bubbler
at 296 K using Ar as a carrier gas at 3.7 umol s'1 on the inner
side of the tubular membrane, and ATSB was introduced
through a bubbler at a temperature of 358 K (lower than in
Example 5) using Ar as a carrier gas at 4.2 mol s', while a
flow ofAr at 17.1 umol s'1 was maintained on the shell side of
the reactor assembly. A dilution gas at a rate of 9.2 umol s"1
was used so that the ATSB concentration was decreased to
0.40><10'3 mol m‘3 while the TEOS concentration was still
19.4><10'3 mol m'3, thus giving rise to a lower molar ratio of
ATSB/TEOS of 0.02. The CVD process was conducted for 3
h and 4.5 h. The resulting composite membrane was desig-
nated SA-V. FIG. 7 and Table 6 show the changes of perme-
ation properties at 873 K before and after CVD. The selec-
tivities increased quickly after the deposition. The membrane
had a SiiAl CVD layer deposited on a gamma-alumina
multilayer placed on top of a macroporous alumina support.
CVD preparation condition V was used with molar ratio of
ATSB/TEOS of0.02 and TEOS concentration of 0.01 94 mol
m‘3. After 4.5 h of deposition, the selectivities of H2 over
CH4, CO and CO2 were 940, 700 and 590, respectively at 873
K while the H2 permeance was 16><10'7 mol m"2 s"1 Pa'l,
which was better than those in Examples 4 and 5. FIG. 8
illustrates the permeation properties of the composite mem-
branes at 873 K prepared using different molar ratios of
ATSB/TEOS.
US 7,585,356 B2
13
TABLE 6
Gas Permeation Properties of a Silica-Alumina Composite
Membrane Before and After Dual-Element CVD at 873 K
14
TEOS concentration (9.l><10'3 mol m'3) was used. In this
example the dilute gas flow rate was increased to 25.4 umol
s"1 so that the TEOS concentration was decreased to 9.l><103
mol m"3 while the ATSB/TEOS was still kept at 0.02. The
CVD process was conducted for 3 h and 6 h. The permeation
Multilayer SA.V properties at 873 K before and after CVD are listed in Table 8.
The low TEOS and ATSB concentrations employed in this
Permeation pmpemes SUbStmte 3 h'CVD 4'5 h'CVD case resulted in a slow increase in the selectivity, indicating
Permeance H2 45 X 1075 2.6 X 1077 1.6 X 1077 that a longer deposition time is required for the preparation of
(mol m’2s’lPa’l) CH4 1.6 x 10’5 2.0 x 10’9 1.7 x 10’10 10 a selective composite membrane in comparison to the cases
CO 1-2 X 10’: 1-2 X 10’: 2-3 X 1071: using higher TEOS and ATSB concentrations such as in. . CO2 9.6x10’ 9.5 x 10’ 2.7 x 10’ EXAMPLES 6 and 7.
Selectivrty H2/CH4 2.8 130 940
H2/CO 3.8 22 700
H2/CO2 4.7 27 590 TABLE 8
15 Gas Permeation Properties of a Silica-Alumina Composite
Membrane Before and After Dual-Element CVD at 873 K
EXAMPLE 7
Multilayer SA-VII
This example describes another silica-alumina composite Pemeafion r0 mics substrate 3 h-CVD 6 h-CVD
membrane formed Oil the gamma-alumina multilayer sub- 20 p p
strate described in EXAMPLE 3. CVD Condition VI was Permeance H2 4-5 >< 10’5 2-5 >< 10’6 4-2>< 10’7
used and the resulting composite membrane was designated (“‘01 “172571155 €34 5'2 X 18:: 3': X 18:: i"; X 18::
SA-VI. In this example, the molar ratio ofATSB/TEOS was Selectivity H2/2CH4 ' :8 ' :3 ' :8
kept at 0.02, the same as in EXAMPLE 6, but a lower TEOS H2/co2 4.7 6.6 26
concentration (l3.0><10'3 mol m'3) was used. TEOS was 25
introduced through a bubbler at 296 KusingAr as a carrier gas
at 3.3 umol s'1 on the inner side ofthe tubular membrane, and EXAMPLE 9
ATSB through a bubbler at a temperature of357 K usingAr as
a carrier gas at 4.1 mol 5*, while a flow ofAr at 22.9 umol . . . . . .
s"1 was maintained on the shell side of the reactor assembly. 30 Thls example describes the preparation ofa Sllrca-alumrna
A dilution gas at a rate of 15.5 umol s"1 was used so that the composrte membrane usmg a n0ngraded1ntermed1ate layer
TEOS concentration was decreased to l3.0><10'3 mol m‘3 on top of a c0mmerc1al mesoporous alumina membrane tube
while the ATSB/TEOS was still 0.02. The CVD process was 21.5 the support. The support purchased from FALL Corpora-
conducted for 3 h and 4.5 h. FIG. 9 and Table 7 show the non (Membralox® TI'7Q'25G Membrane Tube, ”1:7 mm
changes of permeation properties at 873 K before and after 35 O.D.:10 mm) had a n0m1nal pore Size of 5 nm. It was cut into
CVD. The membrane had a SiiAl CVD layer deposited on a length 9f 3'4 cm and then c0nnectedw1th non-porous alu-
a gamma-alumina multilayer placed on a macroporous alu- m1na tub1ng at both ends by ceram1c.J01nts..Bef0re CVD, the
mina support. CVD preparation condition VI was used with top surface of the support was modlfied Whh e thlh layer Of
molar ratio ofATSB/TEOS of 0.02 and TEOS Concentration gamma-alumina membrane by the d1p-calc1n1ng procedure
of 0.0130 mol m'3. The selectivities increased quickly with 40 usmg the d1pp1ng solut10nDS40. CVD C0nd1t10nV was used
the deposition time. After 4.5 h ofdeposition, the selectivities for the preparation. The CVD process was conducted for 3 h
of H2 over CH4 and C02 were 940 and 590, respectively, at and the permeatlon propertles at 873 K before and after CVD
873 K, while the H2 permeance was l.6><10'7 mol m—2 5—1 are listed 1n Table 9 Compared w1th the composrte membrane
Pa'l. deposited on graded alumina multilayer substrate (EX-
45 AMPLE 6), this membrane showed lower permeance and
TABLE 7 selectivity although the same CVD condition was used
Gas Permeation Properties of a Silica-Alumina Composite TABLE 9
Membrane Before and After Dual-Element CVD at 873 K
. 50 Gas Permeation Properties of Silica-Alumina Composite
Multilayer SA'VI Membranes Employing a Commercial Mesoporous Alumina
Support Before and After Dual-Element CVD at 873 K
Permeation properties substrate 3 h-CVD 4.5 h-CVD
Permean2cel 1 H2 4'5 X 107: 3'5 X 107: 2'1 X 107:0 Permeation properties C(2ginrhrpbarlesslig:?fi 3 h-SiO27A1203
(molm’ s’ Pa’ ) CH4 1.6 x 10’ 5.4x 10’ 2.4x 10’
CO2 9.6 x 10’6 9.3 x 1079 6.1 x 10’10 55 Permeance H2 3.3 x 10*5 1.0 ><10’7
Selectivity H2/CH4 2.8 65 870 (mol m’2s’lPa’l) CH4 1.3 x 10*5 3.4 x 10*10
Hz/CO2 4-7 38 340 C02 7.3 x 10*6 6.0 x 10*10
Selectivity H2/CH4 2.6 430
H2/CO2 4.5 170
EXAMPLE 8 60
This example describes another silica-alumina composite EXAMPLE 10
membrane formed on the gamma-alumina multilayer sub-
strate described in EXAMPLE 3. CVD Condition VII was This example describes the hydrothermal stability of the
used and the resulting composite membrane was designated 65 silica-alumina composite membranes. For comparison a pure
SA-VII. In this example, the molar ratio ofATSB/TEOS was
kept at 0.02, the same as in EXAMPLES 6 and 7, but a lower
silica membrane was prepared at 873 K on the same graded
gamma-alumina multilayer substrate by using the CVD) 0f
US 7,585,356 B2
15
TEOS with the TEOS concentration of 19.3><10'3 mol m"3 as
described in our US. patent application Ser. No. 10/775,288,
filed Feb. 10, 2004. The stability test procedure was con-
ducted at 873 K by exposure of the freshly prepared mem-
branes to an environment consisting of 16 mol % water and 84
mol % argon gas or 57 mol % water and 43 mol % argon gas
for 130-520 h. First, anAr flow at a rate of 15 mol s"1 was
passed through a bubbler filled with water at 329 K or 358 K
to produce a wet Ar gas flow containing 16 mol % or 57 mol
% water vapor. Then the wet gas flow was introduced on the
tube side (inner, membrane side) through a stainless steel
tube, which was heated by a heating tape. In the meantime,
another, Ar flow (15 mol s'l) was maintained on the outer
shell side (support side). The H2, CH4 and CO2 permeation
properties were measured periodically during the test. Before
taking permeability measurements, the water vapor was shut
offfor about 20 min to dry the membranes under a dryAr flow.
The wet Ar flow was resumed immediately after the per-
meance measurement. FIG. 10 shows the change in H2 per-
meance ofthe pure silica and composite silica-alumina mem-
branes prepared using different molar ratios ofATSB/TEOS
during the exposure to 16 mol % water vapor at 873 K for 130
h. It was found that although all the membranes suffered a
lowering of their permeance during the first 10 h, the com-
posite membranes clearly showed a strong resistance to the
water vapor compared to the pure silica membrane. After 50
h of exposure to water vapor, the H2 permeance of the com-
posite membranes stabilized, while the permeance through
the silica membrane still kept declining.
On the other hand, it is clear that with increasing molar
ratio of ATSB/TEOS, the resulting composite membranes
showed better hydrothermal stability. FIG. 11 compares the
long-term hydrothermal stability between the pure silica
membrane and the composite membrane SA-IV obtained
under the CVD Condition IV with the molar ratio ofATSB/
TEOS of0.03 . After exposure to 16 mol % water vapor at 873
K for 520 h, the H2 permeance of the composite membrane
SA-IV was decreased by 48%, much less than the 94% for the
pure silica membrane. The membrane SA-IV was further
exposed to a higher humidity environment containing 57 mol
% water vapor at 873 K for another 135 h and the decrease in
H2 permeance was found to be less than 5%. After exposure to
water vapor at high temperature, however, the permeances of
the other gases (CH4 and C02) through the composite mem-
branes changed only little, thus leading to the reduction ofthe
H2 selectivities, as listed in Table 10. This likely occurred
because such gases, with molecular size larger than 0.3 nm,
pass through the membrane through defects and the structure
of these defects did not change much during exposure to the
humid atmosphere.
TABLE 10
Change in the Selectivities ofH2 over CH4 and CO2 through the Composite
Membranes during the Exposure to 16
mol % H207Ar at 873 K
Before exposure 30 h-exposure 130 h-exposure
Membrane H2/CH4 H2/CO2 H2/CH4 H2/C02 H2/CH4 H2/CO2
SA-H 16 30 12 20 8.3 14
SA-IV 170 180 47 66 40 68
SA-V 940 590 270 260 130 300
m
15
35
50
55
60
16
EXAMPLE 11
This comparative example describes another silica-alu-
mina composite membrane deposited using CVD Condition
IV on the gamma-alumina substrate described in EXAMPLE
9, except that air was used as the dilution gas. A molar ratio of
ATSB/TEOS of 0.03 was used for this example. The mem-
brane was denoted as SA-IV—OZ. The dual-element CVD
process was conducted for 3 h. Table 11 compares the per-
meation properties at 873 K of the membrane with a mem-
brane prepared at the identical conditions but with argon as
the dilution gas. It is clear that the use of oxygen severely
deteriorates the permeation properties ofthe membrane. This
indicates that the method of Nam, et al [Korean Membr. J. 3
(2001) 69] is not applicable for making membranes at our
conditions.
TABLE 11
Gas Permeation Properties of Silica-Alumina Composite
Membranes Employing a Commercial Mesoporous Alumina
Support Before and After Dual-Element CVD at 873 K
Multilayer SA-IV SA-IV—O2*
Permeation properties substrate 3 h-CVD 3 h-CVD
Permeance H2 4.5 x 10*5 1.2 x 10*6 2.8 x 10*8
(mol m’2s’lPa’l) CH4 1.6 x 10*5 2.9 x 10*7 1.2 x 10*10
C02 9.6 x 10*6 2.0 x 10*7 <8.0 x 10*11
*Air used as dilution gas instead ofArgon
EXAMPLE 12
This example illustrates the unique permeability properties
of the silica-alumina composite membrane of this invention.
A composite membrane deposited on the gamma-alumina
multilayer substrate was prepared using Condition VI as
described in EXAMPLE 7. The permeability of the mem-
brane for He, H2, and Ne was measured and the results are
presented in FIG. 12. The permeability rises with temperature
as expected. The same unusual order of permeability of
He>H2>Ne was found as in the pure silica membrane
described in our US. patent application Ser. No. 10/775,288,
filed Feb. 10, 2004. Although the present invention is not
intended to be limited by any theory or mechanism, the order
and behavior ofthe species can be explained by a mechanism
originally suggested to describe permeation in vitreous
glasses (J. S. Masaryk, R. M. Fulrath, J. Chem. Phys. 1973,
59, 1198), involving jumps between solubility sites rather
than diffusion through pores. The governing equation is:
hz th (Ns/ NA)1 d2 3* 02 AE /RI_ _ , k
6L( h ]( ankT] (SzfllkT) (ethZkT _ erhv’/2kT)2
Q:
In the equation Qrpermeability, L:membrane thickness,
d:jump distance, h:Planck’s constant, m:mass of permeat-
ing species, k:Boltzmann constant, T:absolute temperature,
anumber of solubility sites, NA:Avogadro’s number,
v:jump frequency, AEK:activation energy, R:gas constant.
This permeance model canbe applied to hydrogen in a similar
form taking into account partial loss of rotational freedom at
the doorway sites through a factor ((jh2/8J'cIkT)2 where o is
the symmetry factor (2) for H2 and I is its moment of inertia.
The calculated curves, assuming ajump distance of0.8 nm, fit
the experimental points very well (FIG. 12). The calculated
parameters are summarized in the following Table 12.
US 7,585,356 B2
17
TABLE 12
Calculated Parameters for Silica-Alumina Composite Membrane
Kinetic Weight
Diameter (atomic NS v* E,
Gases (nm) units) (sites m’3) (s’l) (kl mol’l)
He 0.26 4 8.00x1026 9.00 ><10l2 6.10
H2 0.289 2 6.47 x 1026 1.02 ><10l3 14.1
Ne 0.275 20 6.90 x 1026 4.92 x 1012 12.7
The number of solubility sites is larger for the smaller
species, as on the average there will be more sites available to
accommodate smaller sized species. The jump frequencies
are inversely proportional to the molecular weight of the
species, as lighter species vibrate faster in their equilibrium
sites. The size ofthe solubility sites is smaller than 0.3 nm, as
C02, CO and CH4 do not permeate. The results obtained can
be compared to values for silica membranes with the same
jump distance (0.8 nm) reported in ourU.S. patent application
Ser. No. 10/775,288, filed Feb. 10, 2004,
TABLE 13
Calculated Parameters for Pure Silica Membrane
Gases NS (sites m’3) v* (s’l) Ea (kl mol’l)
He 6.79 x 1026 8.59 x 1012 4.07
H2 4.01 x 1026 1.13 ><10l3 8.90
Ne 5.00 x 1026 4.40 x 1012 8.75
Compared to silica membranes with the same jump dis-
tance, the number of solubility sites in the membrane is
slightly larger, implying that the structure ofthe composite is
a little more restrictive. The jump frequencies are similar,
indicating a similar environment as in silica membranes. The
activation energies are higher, also implying that the compos-
ite membrane has a more restrictive structure.
EXAMPLE 13
This example describes the preparation of a silica-titania
composite membrane deposited on the gamma-alumina mul-
tilayer substrate described in EXAMPLE 3. In this example,
the silica source was TEOS vapor and the titania source was
TIP (titanium isopropoxide) vapor. The same dual-element
CVD apparatus for the deposition of silica-alumina mem-
branes (FIG. 5) was employed except that TIP was used
instead ofATSB (aluminum-tri-sec-butoxide) The CVD pro-
cess parameters denoted as Condition VIII in Table 13 were
used, and the resulting silica-titania membrane was desig-
nated composite membrane ST—VIII. TEOS was introduced
through a bubbler at 296 K using Ar as a carrier gas at 3.7
umol s'1 on the inner side of the tubular membrane, and TIP
through a bubbler at 296 K using Ar as a carrier gas at 10.9
umol 5*, while a flow ofAr of 17.2 mo] 5'1 was maintained
on the shell side of the reactor assembly. A dilution gas at a
rate of2.4 umol s"1 was added so that the TEOS concentration
was 19.5><10'3 mol m‘3 and the TIP concentration was 1.94><
10'3 mol m'3, leading to a molar ratio ofTIP/TEOS of0. 10.
The CVD process was conducted for 3 h and the permeation
properties at 873 K before and after CVD are listed in Table
14. After 3 h of deposition, the selectivities of H2 over CH4
and CO2 were 38 and 58, respectively at 873 K with the H2
permeance of2.3><10'7 mol m"2 s"1 Pa'l, which is less selec-
tive compared with the silica-alumina membranes in
Examples 5, 6 and 7.
10
15
25
30
35
40
45
50
55
60
65
18
The hydrothermal stability test was conducted at 873 K by
exposure of the freshly prepared membrane ST—VIII to an
environment consisting of 16 mol % water and 84 mol %
argon gas for 126 h, as described in Example 10 for the
silica-alumina membrane. After 126 h of exposure, the H2
permeance through the silica-titania membrane was
decreased by 17%. In comparison to the reduction of 90% for
the pure silica membrane and 38-68% for the silica-alumina
membranes tested at the same test conditions in Example 10,
the silica-titania membrane showed better hydrothermal sta-
bility than the pure silica membrane and the silica-alumina
membrane.
TABLE 14
CVD Process Parameters for the Preparation
of Silica-Titania Membrane
Cond. VIII Cond. IX
TEOS Bath Temp. (K) 296 296
TIP Bath Temp. (K) 296 296
TEOS Carrier Gas (pmol s’l) 3.7 3.7
TIP Carrier Gas (pmol s’l) 10.9 5.5
Dilute Gas (pmol s’l) 2.4 8.0
Balance Gas (pmol s’l) 17.2 17.1
TEOS concen. x103 (mol m’3) 19.5 19.3
TIP concen. x103 (mol m’3) 1.94 0.97
TIP/TEOS (molar ratio) 0.10 0.05
Deposition Temp. (K) 873 923
TABLE 15
Gas Permeation Properties of a Silica-Titania Composite
Membrane Before and After Dual-Element CVD at 873 K
Multilayer ST—VIII
Permeation properties substrate (3 h-CVD)
Permeance H2 5.4 x 10*5 2.3 x 10*7
(molm’2 s’l Pa’l) CH4 2.3 x 10*5 6.1 x 10*9
C02 1.4 x 10*6 4.0 x 10*9
Selectivity H2/CH4 2.6 38
H2/CO2 4.4 57
EXAMPLE 14
This example describes another silica-titania composite
membrane formed on the gamma-alumina multilayer sub-
strate described in EXAMPLE 3. CVD condition IX listed in
Table 13 was used and the resulting composite membrane was
designed ST—IX. In this example, the TEOS concentration
was around 19><103 mol m‘3, the same as in EXAMPLE 12,
but a lower molar ratio of TIP/TEOS of 0.05 and higher
deposition temperature of 923 K were employed. The CVD
process was conducted for 3 h and 5 h.After 3 h ofCVD at 923
K, the membrane ST—IX showed a H2 permeance of 2.0><10'7
mol m‘2 5'1 Pa'1 at 923 K with the selectivities of H2 over
CH4 and CO2 of 16 and 26, respectively. With another 2 h of
deposition, the H2 permeance decreased but the selectivities
kept unchanged.
The hydrothermal stability of the composite membrane
ST—IX was tested under very harsh conditions with a much
higher concentration of water vapor (75 mol % water and 25
mol % argon gas) and higher temperature (923 K). The mem-
brane was exposed to 75 mol % water vapor at 923 K for 126
h. Significantly, as shown in FIG. 13 the H2 permeance
through the membrane was reduced by around 24% after the
first 10 h, and then was mostly stabilized in the range of
US 7,585,356 B2
19
24-30% during the remaining hours, and maintained a near-
constant permeance in the order of 10'7 mol m‘2 s'1 Pa'l.
TABLE 16
Gas Permeation Properties of a Silica-Titania Composite
Membrane Before and After Dual-Element CVD at 923 K
Multilayer ST—IX
Permeation properties substrate 3 h-CVD 5 h-CVD
Permeance H2 3.2 x 1075 2.0 x 10*7 1.4 x 10*7
(molm’2 s’l Pa’l) CH4 1.3 x 10*5 1.3 x108 8.4 x10’9
co2 7.3 x 10*6 7.6 x10’9 5.7 x10’9
Selectivity H2/CH4 2.5 16 16
H2/CO2 4.4 26 24
In order to extend the standard-CVD preparation method
we developed an alternative method that employs water vapor
to assist and control the decomposition of the precursors. In
this opposing reactants technique (ORT), the water vapor is
made to permeate from the opposite side ofthe membrane, so
that the Si and Ti precursors (TEOS and TIP) do not decom-
pose on the gas-phase. The setup is shown in FIG. 14, which
is a schematic diagram of a multi-element CVD apparatus
operating with opposing reactants for use in the deposition of
a titania-silica layer. The diluting gas stream was mixed with
the TIP andTEOS carrier gas flows and was introduced on the
inner side ofthe support (tube side). A water vapor carrier gas
flow was passed through a bubbler filled with water (273 or
296 K), premixed with the balance gas flow and then intro-
duced on the outer side of the support (shell side). The water
vapor content in the shell side was adjusted by controlling the
water bath temperature and flow rate. The deposition tem-
perature was varied from 773-873 K. Table 17 lists the pro-
cess parameters for the preparation of TiOzisiO2 mem-
branes by this opposing reactants technique using different
H20 concentrations (0.27 and 1.2 mol %). The TEOS con-
centration was fixed at 0.047 mol % and the molar ratio of
TIP/TEOS at 0.065. The TiOzisiO2 membranes prepared at
773 K by the opposing reactants technique using low (0.27
mol %) andhigh (1 .2 mol %) watervapor concentrations were
denoted as ST—L-773 and ST—H-773, respectively.
TABLE 17
Opposing Reactants Technique Assisted CVD Process
Parameters for the Preparation of Titania-Silica
Membranes (CVD temperature: 673 773 or 873 K)
Membrane 065ST—L 065ST—H
TEOS bath temp. (K) 296 296
TIP bath temp. (K) 296 296
Water vapor bath temp. (K) 273 296
TEOS carrier gas (umol s’l) 3.7 3.7
TIP carrier gas (umol s’l) 7.2 7.2
Diluting gas (umol $71) 6.4 6.4
Water vapor carrier gas 9.6 9.6
Balance gas (pmol s’l) 7.7 7.7
TEOS conc. (mol %) 0.047 0.047
TIP conc. (mol %) 0.0031 0.0031
TIP/TEOS (molar ratio) 0.065 0.065
Water vapor conc. (mol %) 0.27 1.22
EXAMPLE 15
FIG. 15 compares changes with deposition time of the
permeation properties through a standard-CVD derived tita-
nia-silica membrane ST—873) and an ORT assisted-CVD
(ORT—CVD) derived membrane ST—L-873 using a low con-
10
15
20
25
30
35
40
45
50
55
60
65
20
centration ofwater vapor (0.27 mol %). These two composite
membranes were prepared at the same temperature (873 K)
and the same molar ratio of TIP/TEOS (0.065). Clearly, the
opposing reactants technique improves the selectivity prop-
erties. After 3 h of deposition, the H2 selectivities over CH4
and CO, were 20 and 29, respectively for the ORT—CVD
derived membrane, while they were 8.2 and 12 for the stan-
dard-CVD derived membrane. The H, permeance through the
ORT—CVD membrane, however, was slightly lower, 1 .3><10'7
vs 3.3><10'7 mol m"2 s"1 Pa'l. Overall, the introduction of
water vapor improves the membrane properties.
FIG. 16 shows the effect of water vapor concentration on
H2 permeance and HZ/CH4 selectivity with the deposition
time of three composite membranes prepared at 773 K by
using a standard CVD and an opposing reactants technique
assisted CVD. The figure displays changes in the permeation
properties through the composite membranes prepared at the
low temperature of 773 K without and with introduction of
low (0.27 mmol %) and high (122 mol %) water vapor con-
centrations. The presence of water vapor at 773 K not only
shortened the deposition time by at least half, but also
improved the selectivity while maintaining the permeance.
After 7 h ofCVD, the membrane ST—L-773 prepared with the
low water vapor concentration showed a HZ/CH4 selectivity
of 17 with a H2 permeance of 2.8><10'7 mol m"2 s"1 Pa"l at
773 K, while it took 16 h for the membrane ST—773) without
introduction of water vapor to obtain a HZ/CH4 selectivity of
14 with a H2 permeance of3.3><10'7 mol m‘2 s'1 Pa'1 at 773
K. The higher water vapor concentration reduced the selec-
tivity, although it had almost no effect on the permeance for
H2. This may have been due to excessive passage ofthe water
to the side of the membrane with the TEOS and TIP compo-
nents resulting in the homogeneous decomposition of those
reactants.
EXAMPLE 16
The opposing reactants technique allows preparation of
membranes at low temperatures. FIG. 17 shows the properties
of a membrane prepared at 673 K. The figure displays the
permeance and selectivity through a titania-silica composite
membrane ST—L-673 prepared by ORT assisted CVD as a
function of deposition time. This membrane was prepared
using a molar ratio of TIP/TEOS of 0.065, and TEOS and
water vapor concentrations of 0.047 mol % and 0.27 mol %,
respectively. With the use of water vapor the permeance for
H2 was reduced more slowly than that for CH4 and C02, thus
leading to a continuous increase in H2 selectivity. After 18 h of
deposition, the HZ/CO2 selectivity was improved from 2.2 to
8.9 and the H2 permeance was as high as 2.0><10'7 mol m"2
s"1 Pa"l at 673 K. Compared with the results ofthe deposition
at 773 and 873 K in FIGS. 16 and 17, much longer time is
required to obtain a selective composite membrane, but the
performance is not enhanced.
EXAMPLE 17
FIG. 18 shows the effect of water vapor on the H2 per-
meance of composite titania-silica membranes prepared at
773 K using standard CVD and ORT—CVD. The membranes
were exposed to 60 mol % water vapor at 673 K for 130 h.
These three membranes showed similar behavior in their the
resistance to water vapor, probably due to the use ofthe same
preparation conditions including deposition temperature and
molar ratio of TIP/TEOS. They suffered the most significant
loss ofpermeance in the initial states, i.e., the first 20 h. After
40 h of exposure to water vapor, the H2 permeance of the
US 7,585,356 B2
21
composite membranes stabilized. The total reduction of H2
permeance for an exposure of 130 h is in the range of50-60%.
For comparison, the H2, permeance of a pure CVD-derived
silica membrane suffered 90% reduction after exposure to 50
mol % water vapor at 673 K for 100 h.
EXAMPLE 18
FIG. 19 shows the effect of the deposition temperature on
H2 permeance during the exposure to 60 mol % H20 at 673 K
for three composite membranes prepared at 673-873 K using
an ORT—assisted CVD. Interestingly, the membrane ST—L-
873 prepared at 873 K showed very strong hydrothermal
stability. After exposure to a stream containing 60 mol %
water vapor at 673 K for over 120 h, the H2 permeance
through ST—L-873 almost did not change. Its lower per-
meance is due to the use ofthe lower permeation temperature
(673 K) in comparison to its deposition temperature (873 K).
EXAMPLE 19
In order to further investigate the hydrothermal stability of
these composite membranes at very harsh conditions, the
membrane ST—923 prepared at 923 K was exposed at 923 K to
a stream containing 75 mol % water vapor. FIG. 20, shows the
changes in H2 permeance during exposure to 75 mol % H20
at 923 K for a composite membrane prepared at 923 K by a
standard CVD with the use of a molar ratio of TIP/TEOS of
0.065. Significantly, as shown in FIG. 20, the H2 permeance
through this membrane was reduced to around 24% in the first
10 h, and then was stabilized in the range of 24-30% for the
remaining 122 h of the test and still kept a permeance in the
order of 10'7 mol m"2 s"1 Pa'l. In comparison a pure silica
membrane at milder conditions of 873 K at 15 mol % water
underwent a drop ofperrneance of90% to 6><10'8 mol m"2 s"1
Pa'l. The addition ofthe titania to the silica resulted in mem-
branes of excellent permeability and stability
Theory
The permeation properties of the membranes can be cal-
culated using theoretical methods. This section presents
results of ab initio calculations of activation energies made
using a model ofthe structure ofthe silica membrane in which
sorption sites are taken to be randomly placed in the silica
structure. On jumping from site to site, the permeating spe-
cies are considered to pass through a single critical ring open-
ing. A hybrid functional of the density functional theory
(DFT) method with a highly accurate basis set is used to
optimize ring structures and the gas species. The silica rings
were considered to have 6 and 7 members HZMSinOn (n:6.7)
with various elemental substitutions including Al, Ti, B, Y,
and Zr, and are shown in FIGS. 21-23. The geometries of the
ring clusters were obtained by energy minimization of the
structures using a DFT method in Gaussian 98. The optimi-
zations were conducted using the Becke3LYP hybrid func-
tional with a 6-311G (2 d, p) polarized basis set.
EXAMPLE 20
Calculated interaction curves for the permeation ofvarious
gases through 6-member and 7-member membranes com-
poses of Si, SiiAl, and SiiTi are shown in FIG. 24. The
curves show that the interaction energy is highest at the center
of the rings, and decreases with distance. The activation
energy is the highest interaction energy and is summarized for
H2, CH4, and C02, in Table 18. It is seen to increase in the
order H2<COZ<CH4 and to be smaller for the 7-membered
20
40
45
55
22
rings than for the 6-membered rings. Both results are reason-
able, as the molecular size order is H2<COZ<CH4, and the
7-membered rings are larger, allowing easier passage of the
species.
TABLE 18
Summary ofActivation Energies for Permeation ofH2, CO2, and CH4
Through Si SiiAl and SiiTi membrane
Activation energy of permeation/kl mol’l
6-membered rings 7-membered rings
Types ofrings H2 CO2 CH4 H2 CO2 CH4
Silica 27.6 104.3 119.4 10.7 18.1 60.9
Silica-alumina 48.6 111.1 142.6 11.7 18.1 76.4
Silica-titania 32.9 94.9 110.6 14.6 25.9 86.5
EXAMPLE 21
Calculated interaction curves for, the permeation of vari-
ous gases through 6-member and 7-member membranes
composes of SiiB, SiiY, and Siin are shown in FIG. 25.
The activation energies are summarized for H2, CH4, and CO2
in Table 19. As before, the activation energy increases in the
order H2<COZ<CH4, and is smaller for the 7-membered rings
than for the 6-membered rings. Again, the results are under-
standable from the size of the permeating molecules and the
size of the rings.
TABLE 19
Summary ofActivation Energies for Permeation
ofH7 CO7 and CH4 Through
SiiB, SiiY, and Activation energy of diffusioka mol’l
Siin membranes 6-membered rings_ 7-membered rings
Types ofrings H2 CH4 CO2 H2 CH4 CO2
Silica-boron 45.1 108.5 118.3 26.0 62.3 83.2
Silica-yttria 28.2 86.8 109.1 8.70 18.3 46.8
Silica-zirconia 26.5 81.9 113.5 9.67 17.4 43.8
These examples show that the elements comprising the mem-
branes can encompass main group elements B, Si, Al, but also
related elements P, Ga, Ge, As, In, Sn, and Sb, as well as the
early transition metale, Zr, but also the related metals Sc, Ti,
V, Nb, La, Hf, Ta.
What is claimed is:
1. A composite membrane useful for the preferential per-
meation of gases, comprising:
an amorphous mixed-element surface layer comprising
silica and at least one oxide ofan element selected from
the group consisting ofB, F, Al, P, Ga, Ge, As, In, Sn, Sb,
Sc, Ti, V, Y, Zr, Nb, La, Hf, and Ta;
optionally, a porous substrate on which said surface layer is
deposited; and
a porous support on which said substrate or mixed-element
surface layer is deposited;
wherein the permeance of the membrane is greater than
1><10'7 mol m‘2 s'1 Pa'1 and the selectivity of H2 over
C0, C02, and CH4 is larger than 100; and
wherein the membrane incurs a drop in permeance of less
than 80% upon exposure to a stream containing 60
mol % water vapor at 673 K for 120 h.
US 7,585,356 B2
23
2. The composite membrane of claim 1 wherein the mem-
brane incurs a drop in permeance of less than 50% upon
exposure to a stream containing 60 mol % water vapor at 673
K for 120 h.
3. The composite membrane ofclaim 1 wherein the porous
substrate comprises multiple layers having different particle
sizes.
4. The composite membrane ofclaim 1 wherein the porous
substrate comprises a material selected from the group con-
sisting ofalumina, silica, titania, magnesia, zirconia, zeolites,
carbon, phosphorus, gallium, germanium, yttria, niobia, lan-
thana, and mixtures thereof.
5. The composite membrane ofclaim 1 wherein the porous
support comprises a material selected from the group consist-
ing of alumina, silica, titania, magnesia, zirconia, zeolites,
carbon, phosphorus, gallium, germanium, yttria, niobia, lan-
thana, stainless steel and combinations thereof.
6. A method for preparing a composite membrane, com-
prising:
a) providing a porous support;
b) optionally, providing a porous alumina substrate on said
support;
c) carrying out simultaneous deposition of mixed precur-
sors of silica and at least one additional oxide from a
mixed flow stream as to deposit an amorphous mixed-
element surface layer comprising silica and at least one
oxide of an element selected from the group consisting
ofB, F, Al, P, Ga, Ge, As, ln, Sn, Sb, Sc, Ti, V,Y, Zr, Nb,
La, Hi, and Ta, such that the permeance ofthe membrane
is greaterthan l><10'7 mol m"2 s"1 Pa'l, the selectivity of
20
24
H2 over C0, C02, and CH4 is greater than 100, and the
membrane incurs a drop in permeance of less than 80%
upon exposure to a stream containing 60 mol % water
vapor at 673 K for 120 h.
7. The method ofclaim 6 wherein step c) is carried out such
that the membrane incurs a drop in permeance of less than
50% upon exposure to a stream containing 60 mol % water
vapor at 673 K for 120 h.
8. The method ofclaim 6 wherein step c) comprises depos-
iting the mixed precursors by chemical vapor deposition.
9. The method ofclaim 6 wherein step c) comprises depos-
iting the mixed precursors by a sol gel method.
10. The method of claim 6 wherein the mixed precursors
include a precursor of silica that is decomposable by thermal
means.
1 1. The method ofclaim 6 wherein at least one ofthe mixed
precursors is selected from the group consisting of tetraethy-
lorthosilicate (TEOS), tetramethylorthosilicate (TMOS), eth-
yltriethoxysilane, silane, chiorosilane, and combinations
thereof.
12. The method ofclaim 6 wherein at least one ofthe mixed
precursors is selected from the group consisting ofaluminum
tri-sec-butoxide (AISB), aluminum tributoxide, aluminum
tri-tertbutoxide, aluminum triethoxide, aluminum chloride,
and combinations thereof.
13. The method ofclaim 6 wherein at least one ofthe mixed
precursors is selected from the group consisting of titanium
isopropoxide (TIP), titanium alkoxides, titanium alkyls, tita-
nium chloride, and combinations thereof.
* * * * *
UNITED STATES PATENT AND TRADEMARK OFFICE
CERTIFICATE OF CORRECTION
PATENT NO. : 7,585,356 B2 Page 1 of 1
APPLICATION NO. : 11/381088
DATED : September 8, 2009
INVENTOR(S) : Oyama et al.
It is certified that error appears in the above-identified patent and that said Letters Patent is hereby corrected as shown below:
On the Title page Item [75] under “Inventors” please change the inventor’s name “Yungeng” Gu to the
following:
Yunfeng Gu
Signed and Sealed this
Fifteenth Day of June, 2010
David J. Kappos
Dzrector ofthe Umteal States Patent and Trademark Ofice
UNITED STATES PATENT AND TRADEMARK OFFICE
CERTIFICATE OF CORRECTION
PATENT NO. : 7,585,356 B2 Page 1 of1
APPLICATION NO. : 11/381088
DATED : September 8, 2009
INVENTOR(S) : Oyama et a1.
It is certified that error appears in the above-identified patent and that said Letters Patent is hereby corrected as shown below:
On the Title Page:
The first or sole Notice should read --
Subject to any disclaimer, the term ofthis patent is extended or adjusted under 35 U.S.C. 154(b)
by 733 days.
Signed and Sealed this
Fourteenth Day of September, 2010
David J. Kappos
Director ofthe United States Patent and Trademark Ojfice